---
title: "Lobster: with Mearra and AI converting a passive CRM into a secure and effective collaborator"
description: "Lobster partnered with Mearra to build a native AI integration layer, eliminating tedious record-keeping friction for enterprise public affairs teams while strictly upholding institutional compliance. This shifts Lobster from a passive record-keeping database into an active teammate."
date: "2026-06-09T00:00:00+00:00"
image: "/uploads/images/1781020861783-lobster_success_stories.jpg"
image_alt: "Person with business attire and feminine features using tablet computer and having relaxed and happy look on their face. On the background another person, with masculine characteristics, holding huge pile of papers and looking very unhappy."
path: "Success stories > Lobster: with Mearra and AI converting a passive CRM into a secure and effective collaborator"
---

\## The Client

\[Lobster\](https://lobster.ist/) is a Public Affairs CRM and lobbying platform built by industry experts for relationship-intensive domains, including public affairs, government relations, lobbying, and policy advocacy. The software serves as the authoritative record for consultancies, NGOs, public-sector teams, and trade associations. It is engineered to manage stakeholder profiles, tracking, and compliance reporting under the Finnish lobbying registry and the EU Transparency Register.

\## The Challenge: Rigid databases and heavy administrative friction

Public affairs, government relations, lobbying, and policy advocacy are relationship-intensive domains where professional success depends on conversations, relationships, and strategy. However, traditional stakeholder management software often introduces significant administrative friction because systems are typically designed around rigid clicks and forms. As a result, public affairs teams spend an excessive amount of time maintaining database records rather than focusing on core strategic advocacy.

For Lobster's users — including consultancies, NGOs, public-sector teams, and trade associations — every interaction routinely generates a queue of manual data-entry tasks. This continuous administrative overhead forces professionals to spend time updating profiles, logging actions, and modifying project details. This friction diverts strategic experts from relationship management, transforming necessary record-keeping into a burdensome administrative task.

> "Lobbyists gather vast amounts of valuable knowledge through their interactions. Yet the demands of routine documentation often mean that the most important insights never make it into the record, because collecting and inputting that data requires a great deal of time and effort. As a result, organisations risk losing the very intelligence that could compound in value over time."
> 
> Maari Fabritius, Head of UX, Lobster

Furthermore, public affairs is a heavily regulated domain with strict statutory reporting requirements, such as the Finnish lobbying registry and the EU Transparency Register. Because Lobster serves as the authoritative record for these compliance frameworks, data integrity is critical. Any AI solution that bypassed platform controls, made undocumented changes, or short-circuited established workflows would represent a severe compliance liability rather than a productivity tool. Consequently, Lobster required a partner capable of introducing secure AI automation without compromising regulatory guardrails or adding infrastructure complexity.

\## The Solution: A native AI layer via Model Context Protocol

To address these challenges, Lobster partnered with Mearra to design and implement a complete Model Context Protocol (MCP) layer covering the platform's entire data model. MCP is an open standard that allows an AI assistant to interact with an external platform as an integrated component of the system.

The technical implementation integrates the open-standard protocol directly into the existing architecture, avoiding infrastructure bloat while enforcing rigorous compliance guardrails:

-   **Monolithic Integration:** Developed using the fast-mcp Ruby gem, the AI layer compiles directly into the core Lobster Rails application. This approach eliminates the need for separate microservices, external data stores, or integration middleware. The AI tools function as first-class components, utilising the same deployment pipelines, code reviews, and operational monitoring as standard features.
-   **Granular Data Access:** The protocol provides structured capabilities across all platform modules, including stakeholder profiles, organisational hierarchies, projects, action logs, network graphs, and activity timelines.
-   **Per-User Authentication:** Every AI connection requires a personal API key linked to an individual Lobster user account. The AI assistant acts strictly on behalf of that authenticated user rather than a generic service account. Access can be terminated instantly by revoking the API key.
-   **Embedded Compliance Rules:** Compliance is enforced directly at the API layer. Canonical records derived from official government registers are locked against modifications. Furthermore, destructive operations, such as deletions, are blocked automatically if active dependencies or attachments exist, returning explicit validation messages to avoid silent errors.
-   **Validation and Provenance:** Every database write executed by the AI passes through identical Rails model validations and callback chains as human actions, preventing data malformation. The system records the exact same audit trail and user provenance regardless of whether an action was initiated via the web interface or an AI protocol instruction.

Before MCP

With MCP

Create a stakeholder with basic fields only

Create with full profile: contact info, address, social URLs, biography, primary membership with title, dates, and key-position flag

Update notes — nothing else on the profile

Update any editable field on a stakeholder record

No organisation membership management via AI

Add, change, or remove organisation memberships — including switching the primary organisation — in a single request

Cannot delete stakeholders at all

Safe delete with clear error messages specifying what blocks removal

Organisation search broken (server error on every query)

Organisation search fully functional

\## The Results: From manual record-keeping to zero-friction workflows

The full MCP integration redefines how enterprise teams leverage their data assets, transforming Lobster from a record-keeping system into an active strategic asset. By establishing a single data layer shared equally by human users and AI assistants, the technical engagement successfully achieved 100% data accessibility for AI tools. This comprehensive coverage completely closes previous functionality gaps in stakeholder and organisational record management.

> "The Lobster MCP integration is a game changer. Routine documentation, project summaries, reports — all dramatically faster. But the real revelation is being able to combine Lobster's advocacy data with any external source through the same AI agent I already use and love. Instead of building yet another locked-in AI agent like so many SaaS products are rushing to do, Lobster lets you bring the best knowledge and information in the world to your advocacy work. That's not an upgrade. That's a different game entirely."
> 
> Maari Fabritius, Head of UX, Lobster

This architectural shift introduces balanced operational benefits across daily enterprise workflows:

-   **Streamlined Workflows:** Users can execute complex database updates via single conversational instructions. The AI assistant automatically processes multi-step database updates behind the scenes — eliminating manual interface navigation — while maintaining absolute user attribution and security within the audit trail.
-   **Bidirectional Briefing Loops:** Lobster's briefing capabilities now operate as a continuous, bidirectional loop. The AI seamlessly retrieves historical context to compile detailed briefs before major stakeholder interactions and automatically updates relevant records afterward, keeping client files accurate without manual effort.
-   **Rapid Network Traversal:** Teams can unlock latent data value by querying complex stakeholder relationships and historical connections across the platform's network graph. Queries regarding cross-organisational histories that previously required extensive manual cross-referencing across dozens of individual records are resolved by the AI in seconds.

Human user

AI assistant

Works in the Lobster web, or mobile interface or in collaboration with the AI assistant. Handles strategy, relationship decisions, sensitive conversations, and anything that requires judgement.

Works through MCP in collaboration with the human user. Handles data entry, record updates, briefing preparation, and network queries — always acting as the authenticated user.

Crucially for enterprise governance, the implementation remains entirely AI-agnostic. Enterprise clients can securely connect the specific, approved AI tools they already trust — such as Claude or Copilot — via their personal API keys. This strategic flexibility preserves existing corporate IT investments, avoids vendor lock-in, and ensures perfect alignment with internal procurement and security policies.

Please note that there is overlapping ownership between Mearra and Lobster, with certain shareholders holding interests in both companies.

\## If you are interested in friction-free AI workflows, absolute regulatory compliance, and modern public affairs technology, talk to our Vesa or click the send us a message button and we will be in touch very soon.